In a phone call at one o’clock in the morning on March 17, 2003, the U.S. Ambassador to the IAEA, Kenneth Brill, advised IAEA Director General Mohamed ElBaradei to remove his inspectors from Baghdad immediately. The following day, the IAEA gave orders for personnel to leave Iraq. On March 19, the U.S. launched Operation Iraqi Freedom.

Fast forward nine years. We’re now moving into the fifth month of Iran-P5+1 diplomacy without any progress, Prime Minister Netanyahu is urging the powers to declare negotiations a failure, and the drums of war are once again beating in Jerusalem. So it’s no surprise we’re closing out the summer–and for good reason–by revisiting all the potential downsides of an Israeli attack against Iran’s nuclear installations.

Until now, one little item on that list has gotten scarce attention outside the classified world: the messy diplomatic situation Israel would encounter if any IAEA personnel were to be casualties of an air strike on Iran. (It must also be said that the same dilemma would confront the U.S. should, as this account suggested last week, Washington in the more distant future would react to a serious Iranian escalation by taking matters into its own hands).

Might IAEA personnel potentially be at risk in Iran should Israel or the U.S. bomb Iran’s nuclear sites?

According to the current situation on the ground in Iran and what the fine print of the IAEA’s inspection protocols permits the agency to do in the field, without the IAEA having advance guidance or knowledge of whether a military incursion will take place at any specific time, the answer is, in theory, yes.

Iran’s Subsidiary Arrangements

There are IAEA safeguards personnel in Iran 24/7 /365. They are there to carry out safeguards inspections at 16 declared facilities plus if deemed necessary at nine hospitals in Iran which hold nuclear material. The 16 facilities include at least three places I assume would be prime targets of an Israeli air attach in Iran: Natanz, which hosts the Pilot Fuel Enrichment Plant (PFEP) and the Fuel Enrichment Plant (FEP); the Fordow Fuel Enrichment Plant (FFEP); and Esfahan, home to Iran’s Uranium Conversion Plant (UCF).

The IAEA spends quite a bit of time visiting all three sites. Inspectors are not in the plants all the time but they enter them frequently and routinely. At Fordow and Natanz, the IAEA carries out two kinds of inspections: “announced inspections” and “short-notice announced inspections.” At Esfahan and all other sites the IAEA carries out only “announced inspections.”

Meaning what, exactly?

Well, the model protocol for NPT parties’ comprehensive safeguards agreements including Iran’s, Infcirc/153, says this in paragraph 83 under the rubric “Notice of Inspections”:

"The Agency shall give advance notice to the State before arrival of inspectors at facilities or material balance areas outside facilities…(c) for routine inspections… at least 24 hours in respect with the facilities referred to under sub-paragraph 80(b) and sealed stores containing plutonium and uranium enriched to more than 5%, and one week in other cases."

Sub-paragraph 80(b) refers to facilities “with plutonium and uranium enriched to over 5%.” So for the three enrichment plants where IAEA inspectors are currently putting in most of their time and effort, any routine announced inspections would have to be notified by the IAEA one week before the inspection takes place, with the exception of inspections the IAEA would carry out related to the verification of Iran’s enrichment of Iran to 20% U-235 at PFPF and at Fordow, where the at least 24-hour notification rule would obtain.

What about the “short-notice announced inspections” that the IAEA may carry out at Natanz and Fordow? For these we should start with the IAEA’s Safeguards Glossary, where in paragraph 11.7 we’re informed that a “short-notice inspection” is “an inspection performed at a facility… for which less advance notice is provided by the IAEA to the State than that provided for under Infcirc/153.”

That doesn’t tell us how much advance notification the IAEA would have to give Iran, however, because the details of how the IAEA must proceed would follow from confidential subsidiary arrangements which are part of each country’s comprehensive safeguards agreement. Olli Heinonen, who was in charge of the IAEA’s inspection effort in Iran from 2005 until 2010, said here that, generally, ”short-notice” means “typically about two hours.” The details of individual safeguards agreements vary. But I have it on good authority that Iran’s subsidiary arrangements in fact permit the IAEA to conduct a short-notice inspection upon two-hours notice.

Were the IAEA to want to carry out a short-notice inspection at Fordow or Natanz, it wouldn’t tip its hand in advance and would want to be prepared. That would imply that, in order to reap the advantage of surprise, inspectors would have to be pretty close to the enrichment plants at the time Iran would be informed for the inspection to have any value. If inspectors were, say, in Tehran, it would take several hours for them to truck out to Natanz. The farther away they are, the less surprise.

We might conjecture that, absent perfect foresight or guidance, if the IAEA takes seriously the threat that Israel sometime during the rest of this year might launch surprise air strikes against Iran’s enrichment plants, it would not want to put its inspectors at risk by requesting from Iran short-notice inspections at those installations. If the IAEA were to undertake an inspection at Fordow or Natanz, it might move safeguards personnel into locations which were about to be bombed in an air strike the IAEA didn’t know was about to happen.

How Much Advance Warning?

Take a look at the map of three possible Israeli incursion routes into Iran which Paul-Anton Krueger published here. His suggestion that there might have to be several waves of attacks for an attack to succeed, has more recently been taken up in considerable detail by Richard Silverstein, who claims to have been informed about how a complex attack on Iran would unfold. I find it hard to believe that Silverstein has documentary evidence from IDF about how an Israeli attack would proceed. But if Silverstein is generally correct that the attack would proceed in a complex manner, a certain amount of time might elapse between an initial incursion into Iran, meant to paralyze Iran’s electrical distribution system and its communications infrastructure, and an air strike on Iran’s enrichment plants. How much time? Only the architects of such an invasion plan would know. And it would be a fair guess that IDF planners are not talking to any outsiders.

Some observers believe that the ”wave” theory has merit. Ronen Bergman’s account of Iran-Israeli covert warfare suggests that that is how Israel might proceed. In 2007, when Israel carried out an attack against an alleged reactor at Al-Kibar in Syria, according to some people, in addition to seven attack aircraft, it started with unmanned aerial vehicles to disenable infrastructure. Against Iran, Israel would have a number of options. Ballistic missiles, UAVs, cruise missiles, and special operations might be deployed in a battlefield array. It feels safe to conjecture in any case that an attack on Iran would not be the kind of “surgical” strike that Israel carried out in destroying the Osirak reactor in Iraq in 1981.

Then there is the detailed analysis provided by Anthony Cordesman and Abdullah Toukan at CSIS. If an Israeli attack were to be carried out against a large number of targets beyond Iran’s nuclear sites, as they suggest, the scale of operations might be drawn out over a period of several days or even weeks. If that were to happen, any IAEA personnel in the country might be able to conclude in advance that an attack on Fordow or Natanz was imminent because their electric power was cut off and their computers didn’t work. If on the other hand, Israel were to launch Jericho-3 ballistic missiles against Iran’s enrichment plants, a prospect discussed by Cordesman and Toukan, without having advance warning IAEA personnel could be in line of fire.

Iran and Iraq

So to keep IAEA personnel out of harm’s way, would the U.S. or Israel in advance of launching strikes against Iran, as Brill did in 2003, dial up IAEA Director General Yukiya Amano and tell him that he would be well advised to move his inspect0rs out? A lot of people seem to believe there would be a discreet conversation at some point.

Powers attacking Iran would want to remove IAEA personnel including for the reason that, if they didn’t, they would thereafter be excoriated by member states of the IAEA and the U.N. for exposing international civil servants to a surprise attack. But if the attackers intended to keep Iran in the dark, they would have to consider that if they informed the IAEA of their plans, a subsequent exodus of IAEA personel from Iran might signal to Iran that an attack was immanent.

It’s inconceivable that the IAEA would not do everything possible to get personnel out of Iran prior to a surprise attack on Iran’s nuclear sites. But the IAEA must be careful in going about it. If after such an attack information were to leak, or if Amano were compelled to reveal that he had been warned by surprise attackers to withdraw his inspectors, and if the IAEA had chosen not to pass that warning on to Iran, Iran might conclude afterwards that the IAEA was party to an invasion of Iran. Any IAEA personnel still in the country would be at severe risk. If IAEA personnel were out of the country, there would be no direct repercussions, but the IAEA’s relationship with Iran would be over.

Should the IAEA in coming weeks or months, or at any future time decide on its own to avoid these dilemmas by pulling out all safeguards personnel from the country in view of escalating tension between Iran and its adversaries, that step would challenge the IAEA to maintain continuity of knowledge in Iran, as Iran’s uranium enrichment program and other activities remain on autopilot.

Been There Done That

Prior to March 19, 2003, the U.S. had been steadily building up to prepare its invasion of Iraq for months. It wasn’t a secret. I recall preparations being broadcast in real time by reporters on U.S. warships steaming toward the Middle East. I’ve been assuming that the timing of any attack on Iran’s nuclear facilities would remain a deep secret and probably not be signaled in advance by drum-rolls warning Iran to cease and desist or else get bombed. Maybe that assumption is wrong. Were a determined adversary to secretly give Iran an ultimatum to, say, suspend enrichment inside 30 days or face military consequences, and at the same time inform the IAEA that Iran had been so warned, perhaps the IAEA could then judiciously remove its personnel from Iran, and thereafter breathe a sigh of relief, while the Supreme Leader mulls his response.

But Israel would not take risks in a military campaign for the sake of making life easier for the IAEA’s relationship with Iran. So how to get inspectors out of the way of flying bombs without compromising any of the parties involved?

The Vienna missions, their capitals, and the IAEA secretariat, I’m told, know how to do it. What’s more, it has been done before: in Iraq, in the former Yugoslavia, and maybe even previously in Iran.

In any event, the prospect of having personnel in the line of fire is unnerving to the IAEA and its member states. In March 2005, Iran presented to EU negotiators this proposal under the rubric “Elements of Objective Guarantees.” Note item 4 b. “Continuous Presence of On-Site IAEA Inspectors, which can include EU-3/EU nationals, at the UCF and Natanz.” Some people in Vienna will tell you that this item made some diplomats at the outset a little wary: Did Iran’s interlocutors really want to make a blanket commitment to Iran to station personnel 24/7/365 at installations which were increasingly becoming a flashpoint of international tension?

This article was originally published in Arms Control Wonk.